Intelligent Computing Based on Neural Network Model in Problems of Kinematics and Control of Parallel Robot

Abstract

Implementation of a closed-loop control system for a parallel robot is rather difficult due to impossibility of a mechanism output link position direct detection. In order to overcome this, a direct kinematics problem should be solved according to states of the mechanism drive links. However, such approach application requires a solution of a system of nonlinear equations. It is even more complicated for mechanisms with ambiguity of such solution. In this paper, we propose a method to determine the “correct” solution of the direct kinematics problem for such mechanisms using a tripod, a robot with three degrees of freedom, as an example. The method is based on artificial neural networks usage. The data of the mobile platform previous position are used as networks input information. In this paper, it is shown that in this case the adequacy of the obtained neural network based model is sufficiently improved since the neighborhood is defined, where the “correct” solution is located. As a result, both the reliability of the output link position determination and positioning speed and accuracy are improved. The obtained neural network based model can be directly used for the tripod control, and the method to develop this model can be used to synthesize models of other parallel robots.